Two new AI models dropped in 24 hours, Australia just criminalized deepfakes, and Anthropic's IPO countdown has a date.
OpenAI releases GPT-6.1 Terra — and Liquid AI ships LFM3 the same week
OpenAI quietly launched GPT-6.1 Terra on October 26, now available to developers via OpenRouter. Within a day, Liquid AI released LFM3, continuing its push to challenge transformer-based models with its own architecture. These are not headline consumer apps — they are foundation models that other products get built on top of.
Why it matters: The cadence of model releases is accelerating. For anyone building with AI tools at work or school, the underlying models improving means the apps you use get better without you doing anything. But it also means anyone marketing "our AI" without saying which model is probably just reselling someone else's.
Anthropic's IPO roadshow starts November 9 — trading expected before Thanksgiving
Bloomberg reported that Anthropic — the company behind Claude — is expected to begin formal IPO marketing the week of November 9, with trading likely before November 26. The company filed confidentially with the SEC in June and held an investor day in October. It was last valued at $965 billion after a $65B Series H in May.
Why it matters: An Anthropic IPO would be one of the biggest tech listings in years. It is also a signal that AI companies believe now is the moment to go public — either they are confident in the growth story, or they want to lock in the valuation before sentiment shifts.
Queensland makes deepfakes a crime. The rest of the world is watching.
Australia's Queensland state introduced legislation this week making it a criminal offence to create non-consensual sexually explicit deepfake images or audio. The proposed maximum sentence is three years in jail. This comes as the US FTC is also enforcing against platforms under the Take It Down Act for failing to remove harmful AI-generated content.
Why it matters: Laws that were unimaginable two years ago are passing now. Governments are no longer drafting policy papers about deepfakes — they are making arrests and fining platforms. If you have ever thought this only affects celebrities, it does not.
Tenstorrent raises $1.5B for AI chips — the hardware bet is still very much on
Tenstorrent, which makes AI processors and chips as an alternative to Nvidia, closed a $1.5B Series E at a $6.2B valuation, led by Trace Capital and Innovation X. Separately, Hyperbolic — an AI infrastructure startup — raised $30M in a Series A-1 led by Walden Catalyst. The message from investors: compute infrastructure is still the safest bet in AI.
Why it matters: Nvidia does not have the hardware race to itself anymore. Tenstorrent reaching $6.2B means there is real money going into building alternatives. That competition is ultimately what keeps the cost of AI accessible.
India's government tenders for 10,000 GPUs. Startups and universities can apply.
MeitY, through CDAC, has issued a tender inviting AI compute providers to supply 10,000 GPU or TPU units under the IndiaAI Mission. The compute is intended for government ministries, research institutions, startups, and academia. Bidders must offer at least 512 units at a single Indian data centre. This follows MeitY's recent partnership with IBM to accelerate deep-tech startups, especially in Tier 2 and Tier 3 cities.
Why it matters: This is how national AI infrastructure actually happens — a government tender that opens the door to companies and researchers who could not afford cloud compute at market rates. If you are building an AI product in India, this is worth tracking directly.
OpenAI adds election safeguards ahead of US midterms — is it enough?
OpenAI updated its safeguards page this week with new measures covering AI-generated political content, transparency around synthetic media, and systems to monitor political bias ahead of the US midterm elections. This is the first election cycle where AI-generated content at scale is a live, mainstream concern — not a hypothetical. Whether platform-level safeguards keep pace with what the tools can now produce is the question that will define this election season.
Kimi K3.1 Pro open weights due November 4 + Live2.ai raises Rs 130 crore in India
Moonshot AI says the open weights for Kimi K3.1 Pro will be released on November 4, giving developers and researchers free access to a competitive frontier model. In India, Live2.ai raised roughly Rs 130 crore (~$14M) in a pre-Series B round led by IvyCap Ventures — one of the larger recent Indian AI startup rounds. Both are worth a bookmark if you follow the open-source or India startup scene.